Emerging digital trends for journalism

Emerging digital, social, distribution,
hardware, internet and data trends for the
news ecosystem in the coming year.
2018
Tech Trends
For Journalism and Media

F
uture historians will look back on this time as the
turning point for media, information and technol-
ogy. Journalists found themselves in the strange
position of reminding the public the difference
between facts, “alternative facts,” and outright
lies—and then having to defend their centuries-old
commitment to data-driven and research-based reporting. New
tools—from artificial narrow intelligence, to voice interfaces, to
adversarial images—promised to both supercharge newsrooms
and decimate revenue. The usual forms of distribution, for which
there were established business models and profit centers, were
being disrupted by machine learning and gatekeeper algo-
rithms.
Due in large part to technology, the world seemed upside down.
Many journalists were left disoriented, unsure of the path ahead.
Those not feeling extremely uneasy about the future of news
haven’t been paying attention.
There is still time to chart a different course. Buckminster Fuller
once said that “you never change things by fighting the existing
reality.” After all, reality is always in flux—the future is continu-
ally on its heels. “To change something, build a new model that
makes the existing model obsolete,” Fuller said. If news organi-
zations are to survive in the future, they don’t need to make the
existing tenants of quality journalism obsolete. However they
do need to anticipate technological disruption, and prepare for
second, third, fourth, and fifth-order impacts of emerging tech-
nology on the industry. They must develop new models for re-
porting and disseminating the news in order to ensure the long-
term sustainability of operations.
Those in the news ecosystem should factor the trends in this re-
port into their strategic thinking for the coming year, and adjust
their planning, operations and business models accordingly.
The Future Today Institute has published an annual tech trends
report for the past ten years, always focusing on mid- to late-
stage emerging technologies that are on a growth trajectory.
Given all the disruption in news, the timing seemed ripe for a
tech trends report specifically for the future of journalism. This
is the Institute’s first industry-specific report, and it follows the
same approach as our popular annual trends report, which has
now received more than 6 million cumulative views. It is being
released along with our new Global Survey On Journalism’s Fu-
tures, which reveals how those working within journalism think
about the future.
While the trends in this report should help guide your thinking
in 2018, remember that the future never shows up, fully pro-
duced. It is yours to write.
Future historians will look back on this time
as the turning point for media, information
and technology.
Amy Webb
Founder
Future Today Institute
© 2017-2018 Future Today Institute 02

04 Executive Summary
05 Methodology
09 Making Use Of Trends In
Your Organization
13 Artificial Intelligence:
A Primer For Journalists
17 Real-Time Machine Learning
17 Machine Reading
Comprehension
17 Natural Language
Understanding
17 Natural Language
Generation
18 Generative Algorithms
For Voice, Sound and Video
18 Image Completion
Predictive Machine Vision
19 Algorithm Marketplaces
19 Consolidation in AI
20 Uncovering Hidden Bias
in AI
21 Computational Journalism
22 I-Teams For Algorithms
and Data
23 Crowdlearning
24 Adversarial Machine
Learning
25 Computational Photography
26 Bots
27 The Botness Scale
28 Voice Interfaces
29 Ambient Interfaces
30 Deep Linking
31 Productivity Bots
32 Adaptive Learning
33 Nanodegrees
34 Proximity News
35 Personality Recognition
and Analytics
36 Attention
37 Digital Frailty
39 Radical Transparency
Limited-Edition
40 News Products
41 One-To-Few Publishing
42 Notification Layer
43 Journalism as a Service
44 Transparency in Metrics
45 Real-Time Fact Checking
46 Offline Is The New Online
47 Audio Search Engines
48 CubeSats
50 Connected TVs
51 WebRTC
51 Streaming Social Video
52 New Video and Audio
Story Formats
53 Splinternets
54 Media Consolidation
56 Blocking the Ad Blockers
57 Natural Language
Generation for Reading
Levels
58 Leaking
59 The First Amendment in
a Digital Age
60 Personal Networks
61 Holograms
61 Virtual Reality
62 360-degree Video
62 Augmented Reality
63 Mini-Glossary of Mixed
Reality Terms For Journalists
65 Differential Privacy
65 Trolls
66 Authenticity
66 Data Retention Policies
66 Backdoors
67 Prize Hacks
67 Weaponizing Wikileaks
67 Glitches
68 Ownership
69 Hacker Terms and Lingo
Every Journalist Should
Know
75 Organizational Doxing
76 Blockchain For Journalism
77 Sense And Avoid Technology
77 Drone Swarms
77 Drone Lanes
77 Clandestine, Disappearing
Drones
78 Autonomous Underwater
Vehicles
78 Microdrones
78 Drone Delivery
79 Head Mounted Displays
79 Smartwatches
80 Earables
80 Thinkables
81 Internet of X
82 5G
83 About The Future Today
Institute
84 About The Author
84 Special Thanks
85 Disclaimer
87 Company Index
91 Contact Information
Table of Contents

© 2017-2018 Future Today Institute
Executive Summary
The Future Today Institute’s 2018 Tech Trends For Journalism Report is our first industry-specific
analysis of emerging technology trends. It follows the same approach as our popular annual trends
report, now in its 10th year of publication with more than 6 million cumulative views.
• In 2018, a critical mass of emerging technologies will converge,
finding advanced uses beyond initial testing and applied re-
search. That’s a signal worth paying attention to. News organi-
zations should devote attention to emerging trends in block-
chain, voice interfaces, the decentralization of content, mixed
reality, new types of search, and hardware (such as CubeSats
and smart cameras).
• Journalists need to understand what artificial intelligence is,
what it is not, and what it means for the future of news. AI re-
search has advanced enough that it is now a core component
of our work at FTI. You will see the AI ecosystem represented
in many of the trends in this report, and it is vitally important
that all decision-makers within news organizations familiarize
themselves with the current and emerging AI landscapes. We
have included an AI Primer For Journalists in our Trend Report
this year to aid in that effort.
• Decentralization emerged as a key theme for 2018. Among the
companies and organizations FTI covers, we discovered a new
emphasis on restricted peer-to-peer networks that detect ha-
rassment, share resources and connect reporters with sources.
There is also a push by some democratic governments around
the world to divide internet access and to restrict certain con-
tent, effectively creating dozens of “splinternets.”
• Consolidation is also a key theme for 2018. News brands,
broadcast spectrum, and artificial intelligence startups will
continue to be merged with and acquired by relatively few
corporations. Pending legislation and policy in the U.S., E.U.
and parts of Asia could further concentrate the power among
a small cadre of information and technology organizations in
the year ahead.
• To understand the future of news, you must pay attention to
the future of many industries and research areas in the com-
ing year. When journalists think about the future, they should
broaden the usual scope to consider developments from myr-
iad other fields also participating in the knowledge economy.
Technology begets technology. We are witnessing an explo-
sion in slow motion.
04

Methodology
The Future Today Institute’s
forecasting model relies on
quantitative and qualitative data.
Our model alternates between
flared and focused thinking. This
includes: identifying very early
stage fringe research, focusing
on patterns, interrogating trend
candidates, calculating a trend’s
trajectory, writing scenarios and
finally pressure-testing strategies
and recommendations.
Forecasting Methodology: The Six-Step Funnel
2
3
4
6
5
1
The fringe
CIPHER
Ask the right questions
Calculate the ETA
Write scenarios
Pressure-test
the future
Answers
Make observations and harness
information from the fringes of
society or a particular research
area.
Uncover hidden patterns
by categorizing information
from the fringe: contradictions,
infections, practices, hacks, extremes,
rarities.
Ask the right questions to determine
whether a pattern is really a trend.
Ensure that the timing is right for
the trend and for your organiztion.
Scenarios inform the strategy you
will create to take the necessary
action on a trend.
Are your scenarios comprehensive
enough? Is your level of confidence
justified? Is the strategy you’re
taking the right one for the future?
What is the future of X?

How To Use The 2018 Tech Trends
For Journalism Report
Our 2018 Trend Report reveals strategic opportu-
nities and challenges for your news organization
in the coming year.
The Future Today Institute’s first-ever Tech Trends For Journalism
and Media Report prepares staff, managers, executives, funders
and startups for the year ahead, so that they are better posi-
tioned to see technological disruption before it fully erupts. Use
our report to identify near-future business disruption and com-
petitive threats while simultaneously finding new collaborators
and partners. Most importantly, use our report as a jumping off
point for deeper strategic planning.
Explaining why these trends matter.
Rather than simply offering an overview of the trends that will
matter in 2018, this report takes the additional step of explain-
ing why and how these trends will impact your organization. In
some cases, you will see very specific use cases and descriptive
illustrations, so that you can more clearly envision the potential
outcomes of these trends during the next 12 months.

How To Use Our Report
Each trend offers six important pieces
of information for newsrooms.
01 Key Insight
Short, easy explanation of this trend so that you
can internalize it and discuss with your colleagues.
02 Examples
Real-world use cases, some of which will sound
familiar.
03 What’s Next
What this trend means for you and your news
organization in the coming year.
04 Watchlist
Notable companies, founders and researchers
working in this trend space.
05 Years on the List
We’ve noted how many years we’ve been tracking
the trend in our annual Tech Trends Report, which
began publication 10 years ago. This measurement
is an indication of how the trend is progressing.
06 Action Matrix
An easy-to-read graphic indicating whether the
trend needs monitoring, should inform your
strategy, or requires action.
01
05
0
03
06
04

10 Questions
How to relate these trends back to your news organization in 2018
Our Tech Trends For Journalism and Media Report is meant to inspire you to generate new ideas. Use
it as you listen for the signals talking and to advise your strategic thinking throughout 2018.
As you think about the trends in this report, ask your team and yourself the following questions:
01 How might this trend impact the news industry
and all of its parts?
02 What are the second, third, fourth, and fifth-order
implications of this trend, both on my newsroom
and on our industry?
03 Does this trend signal greater disruption to our
traditional business practices and subscription
models?
04 Does this trend indicate a future disruption to
established roles and responsibilities within our
organization? If so, how can we reverse-engineer
that disruption and deal with it in the present day?
05 How are companies/ agencies/ organizations in
adjacent spaces–outside of news–addressing this
trend? What can we learn from their best practices?
06 How are our competitors/ related agencies
harnessing this trend (or failing to do so?)
07 How will the wants, needs and expectations of
our customers change as a result of this trend?
08 How does this trend inspire me to think about
the future of news and my role within the news
ecosystem?
09 How does this trend inspire my team/
organization?
10 How does this trend help me/ my team/
my organization think about innovation?

How To Take Action
On Tech Trends
One of the most difficult challenges
organizations must confront is a will-
ingness to take incremental action.
Many organizations prefer to “wait
and see” before taking action. How-
ever, it’s precisely that waiting which
causes companies to fall behind and
miss opportunities.
The Future Today Institute uses a sim-
ple framework to continually monitor
technology as it moves from fringe to
mainstream. Incremental actions po-
sition a business unit to make smart-
er strategic decisions when the time
is right.
Below is our framework, and we en-
courage your organization to use it
for creating incremental action on
tech trends.
“Canwedoit”
Learning Stage
As we research and test this
new technology, what can
we learn and apply to our
organization?
Listening for Signals
Emergent ideas at the
Fringe, experiemntation
and trials from the “unusual
suspects” both outside and
inside your industry
Watching The Horizon
Emerging but bona-fide
technology and trends;
uncertain trajectory and
timeline; ecosystem
forming; market forming
Developing Ideas Stage
Can we develop a new
product or service that
leverages the technology,
even as it is still evolving?
Capability Building Stage
How can we work to
more fully understand the
emerging technology and
develop the expertise to act?
UncertaintyAboutATechnology
Uncertainty About Technology in the Market
High
Low
Low High
“Does the market want it”
Framework For Incremental Action On Tech Trends

FAQ
What Is A Trend, Exactly?
Mapping the future of the news ecosystem begins
with identifying early signposts as you look out on
the horizon. In order to chart the best way forward,
you must understand emerging trends: what they
are, what they aren’t, and how they operate.
At any moment, there are hundreds of small shifts
in technology—developments on the fringes of
science and society—that will impact our lives in
the future. A trend is a new manifestation of sus-
tained change within an industry sector, society,
or human behavior. A trend is more than the lat-
est shiny object.
Fundamentally, a trend leverages our basic hu-
man needs and desires in a meaningful way, and it
aligns human nature with breakthrough technolo-
gies and inventions.
All trends share a set of conspicuous, universal
features:
• A trend is driven by a basic human need,
one that is catalyzed by new technology.
• A trend is timely, but it persists.
• A trend evolves as it emerges.
• A trend can materialize as a series of uncon-
nectable dots which begin out on the fringe
and move to the mainstream.
Identifying something as a trend means connect-
ing the dots, or relating changes in the present
to what’s coming in the future. To map what the
future holds, seek out the early adopters, the
hackers, the developers with seemingly impossi-
ble ideas. It’s within these circles that meaning-
ful changes begin. As the trend evolves, the work
of these disparate groups begins to overlap, un-
til it converges in a single point—before perhaps
evolving once again.

Technological advancement influences future changes and disruption across fields and industries. If
you hope to understand the future of news, you can’t just look at trends within a silo. To forecast the
future of the news ecosystem, you need to plot out the intersecting vectors of technological change
looking through these ten modern sources of change.
FAQ
Because trends are a different way of seeing and interpreting our current reality, they provide a useful
framework to organize our thinking, especially when we’re hunting for the unknown and trying to learn
something about which we do not yet know how to ask. There are ten modern sources of change in
society with technology as the primary connector.
Wealth
distribution
Education Government Politics Public health
Demography Economy Environment Journalism
Media
(our individual and
collective use of social
networks, chat services,
digital video channels,
photo sharing services
and so on)
01
06
02
07
03
08
04
09
05
10

2018 Tech Trends

Artificial Intelligence
Eighth year on the list
Key Insight
Many facets of artificial intelligence (AI) have made our list
since we first started publishing our FTI report a decade
ago. AI itself isn’t the trend—it is too broad and important
to monitor without distinguishing between signals. For that
reason, we have identified different themes within AI that
you should be following.
AI: A Primer For Journalists
What You Need To Know About AI
Simply put, AI is a branch of computer science in which
computers are programmed to do things that normally re-
quire human intelligence. This includes learning, reasoning,
problem-solving, understanding language and perceiving
a situation or environment. AI is an extremely large, broad
field, which uses its own computer languages and even spe-
cial kinds of computer networks that are modeled on our
human brains.
AI’s History In Brief
The idea that we might someday create artificially intelli-
gent, sentient robots was first suggested by prominent phi-
losophers in the mid-1600s. Mathematician Ada Lovelace,
in the footnotes of a paper she was translating, posited the
theory that someday a computer might be capable of cre-
ative acts—and to think, just like we humans do. Computer
scientist Grace Hopper pushed that idea forward, pioneer-
ing early programming languages that were similar to spo-
ken English. For the past six decades, researchers have been
working towards a functional AI, using the human brain for
inspiration, but they didn’t have access to enough compute
power, data or people trained to advance the field. As a re-
sult, the field entered what’s known as the “AI winter,” when
funding and enthusiasm dried up. In the past decade, new
advances by Google, Amazon, Microsoft, Tencent, Baidu,
Facebook, Apple, IBM and universities around the world
have reignited excitement and funding.
There Are Different Categories Of AI
There are two kinds of AI—weak (or “narrow”) and strong
(or “general”). When Narrative Insights writes a story out
of structured data, that’s ANI. Outside of journalism, there
are hundreds of examples of ANI in everyday life: the spam
filters in your email inbox, the recommendation engines on
Amazon and Netflix, the anti-lock breaks in your car, the
prices you see when you buy air tickets. The H.A.L. super-
computer from 2001: A Space Odyssey, which was not only
sentient, but decided it no longer had use for us humans, is
a representation of artificial general intelligence (AGI).
13
TRENDS 001 - 009

AI, Neural Networks and Deep Neural Networks
A neural network is the place where information is sent and
received, and a program is the set of meticulous, step-by-
step instructions that tell a system precisely what to do so
that it will accomplish a specific task. How you want the
computer to get from start to finish—essentially, a set of
rules—is the “algorithm.”
AI, Machine Learning and Deep Learning
Machine learning programs run on neural networks and
analyze data in order to help computers find new things
without being explicitly programmed where to look. Within
the field of AI, machine learning is useful because it can help
computers to predict and make real-time decisions without
human intervention.
Deep learning is a relatively new branch of machine learn-
ing. Programmers use special deep learning algorithms
alongside a corpus of data—typically many terabytes of
text, images, videos, speech and the like. Often, these sys-
tems are trained to learn on their own. In practical terms,
this means that more and more human processes will be
automated. Including the writing of software, which com-
puters will soon start to do themselves.
14
Artificial Intelligence cont.
TRENDS 001 - 009

AI, Automation and Augmented Journalism
Early experiments at the LA Times and at startups such as
Narrative Science and Arria NLG have proven that AI sys-
tems can transform raw data into narratives, crafting sto-
ries that seem as though they’ve been written by a human.
Earthquakes, sports recaps, financial summaries and crime
reports have now been written by automated systems and
published by news organizations. For now, these systems
are only capable of telling the story of “what” autonomous-
ly. Other AI systems can be used to augment the analytical
thinking of journalists—working alongside these systems,
journalists have a supercharged ability to uncover and un-
derstand the “why.” However in the not-too-distant future,
new generations of these systems will be able to do that
autonomously, too.
Subjective Interpretation
One of the challenges to pushing the limits of automation is
in subjective interpretation: what makes a number “big” or
“small” certainly depends on circumstances. For example, in
the 2016 presidential election, there were times when poll-
sters reported that Hillary Clinton held a 6-point lead over
Donald Trump. In that particular case, a 3-point lead would
have seemed low—a significant detail. On the other hand, if
that had been a Baltimore city mayoral election, a 3-point
lead separating the two frontrunners would have been sta-
tistically important. That’s because the mayoral election
tends to get decided during the primary. Democrats always
win, and by a massive margin.
In our present-day machine learning models, these excep-
tions must be thought out in advance by humans and taught
to machines. That’s not an easy task at the moment.
Automating Journalism
Unfortunately in journalism, AI has become a popular short-
hand for “automation.” AI will not solve all of the problems
with the news media business, and it cannot—at least, not
right now—take the place of trained journalists in a news-
room. The challenge with declaring AI in newsrooms a fait
accompli is that we are only at the very beginning of the
artificial intelligence era.
In the next 24-36 months, computer vision, natural lan-
guage algorithms, generative content algorithms, deep
learning—along with increased compute power, lots of
data and more ubiquitous accessibility to tools—will coa-
lesce and allow journalists to do richer, deeper reporting,
fact checking and editing. Many of the trends that follow,
from machine reading comprehension to predictive ma-
chine vision to computational photography will give jour-
nalists superpowers, if they have the training to use these
emerging systems and tools.
15
TRENDS 001 - 009

Voice Is Journalism’s Next Big Challenge
Also looming on the horizon: voice interfaces, which is a
key component of the future of AI and content. By 2023,
50% of the interactions consumers have with all computers
will be using their voices.1
Think about the implications of
people having conversations with machines. If a consumer
wanted to know the latest information about an election,
she’d ideally just ask: “What’s happening with the elec-
tion? Who’s in the lead?” At that point, the system she’s
talking to would have two options: either choose just one
news source and start a response with “according to the
[news source],” or otherwise pull information from many
sources and have a more robust conversation. However in
that case, how do news organizations get cited for their
reporting? Does the system continually interrupt itself to
say where the news is coming from? That’s now how two
humans would interact with each other.
Once we are speaking to our machines about the news,
what does the business model for journalism look like?
News organizations are ceding this future ecosystem to
outside corporations. They will lose the ability to provide
anything but content. When speaking to machines, con-
sumers may not know which media brand they’re having
a conversation with.
1
This number is based on Future Today Institute modeling and applies only to North America.
While some news organizations have started to experi-
ment with chat apps and voice skills on Alexa and Goog-
le Home, journalism itself is not actively participating in
building the AI ecosystem. News organizations are cus-
tomers, not significant contributors. We recommend
cross-industry collaboration and experimentation on a
grand scale, and we encourage leaders within journalism
to organize quickly. AI does pose an existential threat to
the future of journalism.
16
TRENDS 001 - 009

01 Real-Time Machine Learning
It is recently possible to use a continual flow of transactional data
and adjust models in real-time. Potential use cases include match-
ing news consumers to the right product as they are looking at a
website, as well as re-writing content on a site to match the needs
of each individual user. In addition, it promises real-time fraud de-
tection and security measures such as authenticating someone
based on her typing habits.
02 Machine Reading Comprehension (MRC)
For AI researchers, machine reading comprehension has been a
challenging goal, but an important one. If you perform a search
query, wouldn’t you rather have a system offer you a precise an-
swer than just a list of URLs where you can go to hunt down more
specifics—even showing you where, on the page, that informa-
tion comes from? That’s the promise of MRC. MRC isn’t focused
on keywords alone. In the future, a trained MRC system could be
transferred to different domains where no human has created la-
bels or even a standard taxonomy—and the MRC would be able
to read, infer meaning, and immediately deliver answers. MRC is a
necessary step in realizing artificial general intelligence, but in the
near-term it could potentially turn a news organization’s website
into a searchable repository of information. This could be espe-
cially useful once voice-based interfaces become more common.
03 Natural Language Understanding (NLU)
We are surrounded by unstructured text in the real world—it ex-
ists in our social media posts, our blog entries, on company web-
sites, within city hall digital records, and elsewhere. NLU allows
researchers to quantify and learn from all of that text by extract-
ing concepts, mapping relationships and analyzing emotion. NLU
capabilities would allow news organizations to sift through heaps
of documents and gain insights much faster than reporters going
at it alone.
04 Natural Language Generation (NLG)
Algorithms can transfer data into a narrative using natural lan-
guage generation. Dozens of news and other organizations,
including Bloomberg and the Associated Press, are using
Automated Insights, which mines data and is capable of writ-
ing more than 2,000 stories per second. They will use natural
language generation to produce stories about fantasy football,
earnings reports and the like. Narrative Science employs its
NLG system to build narratives out of big data sets and to help
non-data science people make better sense of what’s happen-
ing within their organizations.
17
TRENDS 001 - 009
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
High Degree of Certainty
Low Degree of Certainty
ImmediateImpactontheNewsIndustry
Longer-TermImpactontheNewsIndustry

05 Generative Algorithms For Voice,
Sound and Video
Researchers at MIT’s CSAIL are studying how children learn
new words in order to train computers on automatic speech
recognition. As humans, we are able to master a new concept
from just one or two examples; for machines, this is a more diffi-
cult task when it comes to language. Meanwhile, researchers are
training computers to watch videos and predict corresponding
sounds in our physical world. For example, what sound is gener-
ated when a wooden drumstick taps a couch? A pile of leaves?
A glass windowpane? The focus of this research is to help sys-
tems understand how objects interact with each other in the
physical realm. But future versions of the algorithms could be
used to automatically produce sound and sound effects for
news videos, movies and TV shows. It also raises the specter of
audio fraud—what happens when computers are able to spoof
our voices and natural sound? There have already been a few
early successes: in 2017, researchers at the University of Wash-
ington developed a model that convincingly showed President
Barack Obama giving a speech—that he never actually gave
in real life. Journalists must start applying additional scrutiny
to audio, sound and video obtained from sources outside the
newsroom.
06 Image Completion
If a computer system has access to enough images—millions
and millions—it can patch and fill in holes in pictures. There
are practical applications for journalists—if the foreground of
a mountain is out of focus, another version of the scene can
be swapped in to generate the perfect picture. However, there
are ethical considerations as well. How much image completion
should be allowed? How and when do you draw a line between
reality and enhancement? Image completion is also a useful tool
for law enforcement and military intelligence officers—comput-
ers can now assist them in identifying who or what is in the
frame. Given the bias we’ve already seen across machine learn-
ing algorithms and data sets, image completion could wind up
being a public interest story in the coming years.
18
TRENDS 001 - 009

07 Predictive Machine Vision
Researchers at MIT’s CSAIL have trained computers to not only
recognize what’s in a video, but to predict what humans will do
next. Trained on YouTube videos and TV shows such as “The Of-
fice” and “Desperate Housewives,” a computer system can now
predict whether two people are likely to hug, kiss, shake hands
or slap a high five. This research will someday enable robots to
more easily navigate human environments—and to interact with
us humans by taking cues from our own body language. It will
also help with personalized recommendations—it could usher in
an era of aggressively versioned distribution, where news con-
sumers would see a news experience customized specifically for
them using predictive modeling.
08 Algorithm Marketplaces
Most news organizations can’t staff a team of developers who
have unlimited time to create, test and refine algorithms. As a
result, communities of developers are offering up their algo-
rithms in emerging algorithm marketplaces. Algorithmia is
like Amazon but for algorithms, where developers can upload
their work to the cloud and receive payment when others pay
to access it. DataXu offers a marketplace for its proprietary
algorithms. Quantiacs allows developers to build algorithmic
trading systems, and it matches their algorithms up with capi-
tal from institutional investors. PrecisionHawk has launched a
marketplace for predictive agriculture algos. A number of other
networks, such as Nara Logics, MetaMind, Clarifai offer tools
for developers to build deep learning into any application. Look
for even more niche marketplaces in 2018.
09 Consolidation in AI
Some in the AI ecosystem now worry that the future of AI is
already under the direction of too few companies. Just a hand-
ful of companies dominate the AI landscape: Google, Amazon,
Tencent, Baidu, IBM, Facebook, Apple and Microsoft. On
the investment side, Intel Capital, Google Ventures, GE Ven-
tures, Samsung Ventures, Tencent and In-Q-Tel lead. As with
any technology, when just a few companies dominate the field,
they tend to monopolize both talent and intellectual property.
They’re also partnering to build on each others’ work. When it
comes to the future of AI, we should ask whether consolidation
makes sense for the greater good, and whether competition—
and therefore access—will eventually be hindered as we’ve seen
in other fields such as telecommunications and cable.
19
TRENDS 001 - 009

Key Insight
By now, it’s no secret that some of our machine learning
models—and the data they use—are encoded with bias.
That’s because the people who built the models are them-
selves subject to unconscious bias, as well as more explicit
homogeneous learning and working environments.
Examples
In 2016, ProPublica.org published an exceptional inves-
tigation on machine bias and the problem of using AI to
predict future criminals. Their findings: so-called “risk as-
sessment” software is increasingly common in courtrooms
across the nation, and it is used to inform decisions about
everything from bond amounts to the length of a criminal
sentencing. Perhaps unsurprisingly, the software is biased
against people with darker skin than those with lighter skin.
(We encourage you to read ProPublica’s full report: https://
www.propublica.org/article/machine-bias-risk-assess-
ments-in-criminal-sentencing.)
What’s Next
Risk assessment software is hardly an outlier. Numerous
studies undertaken by prominent universities, including
MIT, Harvard, Carnegie Mellon, Princeton, University of Cal-
ifornia-Berkeley (among others) have shown explicit bias
in algorithms across many industries and social sectors. Al-
gorithmic bias is a problem that will get worse. Computers
are trained using a limited initial set of data, and the training
programs are built by humans. Often, the training sets re-
veal unacknowledged bias hidden within us.
As newsrooms incorporate datasets, machine learning and
computer vision into their reporting, it’s imperative that
journalists learn how to investigate the data itself as well as
the models used to interpret and learn from that data.
Watchlist
Investigative Reporters Editors; National Institute for
Computer-Assisted Reporting; MIT; Harvard University; Car-
negie Mellon University; Stanford University; University of
California-Berkeley; Brown Institute at Columbia University;
Tow Center for Digital Journalism at Columbia University;
Philip Merrill College of Journalism at the University of Mar-
yland; Coral Project; ProPublica.
20
Uncovering Hidden Bias in AI
Third year on the list
TREND 10
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
From ProPublica’s investigative report on “risk assess-
ment” software.

Key Insight
What are the ways in which data and algorithms can en-
hance reporting? Computational Journalism builds on the
25-year foundation of Computer Assisted Reporting (or
CAR) is an investigative journalism technique. Reporters
find, clean and mine public records and documents, crunch
data and uncover hidden stories. Aided by machine learn-
ing algorithms and AI, computational journalism is the evo-
lution of CAR.
Examples
It’s one thing to find and mine public data—analyzing what’s
there, and connecting the seemingly unconnectable dots,
is another challenge entirely. Computational journalism
techniques such as multi-language indexing, automated
reporting, entity extraction, algorithmic visualization, mul-
tidimensional analysis of data sets, flexible data scraping,
are allowing journalists to combine what they find in the
data and then see the connections between facts, keywords
and concepts. In this way, they can reveal interconnected
relationships between people and organizations that they
might not have otherwise seen.
What’s Next
We anticipate increased demand in computational journal-
ism and journalists with complimentary skills sets. There are
a host of stories waiting to be discovered, written and pro-
duced.
Watchlist
Investigative Reporters Editors; National Institute for Com-
puter-Assisted Reporting; Coral Project; Stanford Computa-
tional Journalism Lab; Duke University; University of British
Columbia; University of Texas at Austin; Brown Institute at
Columbia University; Tow Center for Digital Journalism at
Columbia University; Philip Merrill College of Journalism at
the University of Maryland; Media Change and Innovation
Division at the University of Zurich; Annenberg School of
Communication Journalism and the University of South-
ern California; Wall Street Journal; New York Times; Wash-
ington Post; Tamedia; ProPublica; National Public Radio.
21
Computational Journalism
TREND 11
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
IBM’s News Explorer is an example of a computational
system for reporting.

Key Insight
News organizations need a new kind of special-ops team:
investigative reporters who specialize in investigating the
algorithms and data itself.
Examples
Algorithms, data sets and AI systems reflect the worldviews
of their architects and trainers. This information is used to
help make decisions, to predict behavior, and to generate
answers to questions. More of these systems now govern
everyday life and are used by law enforcement, universities,
financial institutions and government agencies. Journalists
must begin to investigate how the data and algorithms in-
tersect with daily life. And, to prevent bias in reporting, jour-
nalists must gain a better understanding of who created the
algorithms and data sets, and what their processes were. For
example, the PredPol predictive policing system, which is
used by police departments around the U.S., recommended
time and time again that departments concentrate their ef-
forts on neighborhoods that were overwhelmingly poor and
black. The problem has to do with how arrest data is gath-
ered, and how individual police departments have historically
monitored their local communities. The model didn’t include
a rigorous check on bias in the initial data sets. Reporters at
the New York Times, Wall Street Journal, ProPublica and
Washington Post have been applying the core practices and
skills of reporting to investigating algorithms.
What’s Next
We will soon reach a point when we will no longer be able
to tell if a data set has been tampered with, either intention-
ally or accidentally. AI systems rely on our trust. If we no
longer trust the outcome, decades of research and techno-
logical advancement will be for naught. Building trust and
accountability is a matter of showing the work performed.
This is a complicated process, as understandably news or-
ganizations would want to keep certain data and reporting
methods private.
Watchlist
Brown Institute at Columbia University; Macromedia Uni-
versity of Applied Sciences; Tow Center for Digital Journal-
ism at Columbia University; AlgorithmWatch.org; ProPubli-
ca; Philip Merrill College of Journalism at the University of
Maryland; Media Change and Innovation Division at the Uni-
versity of Zurich; Annenberg School of Communication
Journalism and the University of Southern California; Wash-
ington Post; New York Times; Wall Street Journal; National
Public Radio; Investigative Reporters Editors; National In-
stitute for Computer-Assisted Reporting.
22
I-Teams For Algorithms and Data
First year on the list
TREND 12
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
AI systems rely on our trust.

Key Insight
You’re familiar with crowdsourcing: asking the public to con-
tribute content or to assist with on-the-ground reporting on
an issue. Crowdlearning is a computational journalism tech-
nique that queries our passive data—our mobile and online
activity, our public health records, our locations—to learn or
understand something new.
Examples
In June 2016, the evening after citizens in the United King-
dom voted for Brexit, Google revealed sobering search
data: people in the UK were Googling “what is the EU.” This
passive data told an interesting story, and it’s just part of
what we’re now able to learn from the crowd by monitoring
various networks. Our smartphone ownership has reached
critical mass, and so has our use of various networks. Our
data not only follows us around, it’s often available for any-
one to search, collect and analyze.
What’s Next
Good crowdlearning sources are already available to us, and
they include HealthData.gov, Google’s busy times data for
businesses and public spaces, Waze, Wikipedia and more.
We anticipate that more news organizations—as well as
marketers, activists and other groups—will start harnessing
data in creative ways. That’s because our thinking results
in behavior (like searching for “what is the EU?”). Our be-
havior results in data. And that data can be used to learn
something about us.
Watchlist
Google; Bing; Apple; Microsoft; Investigative Reporters and
Editors; National Institute for Computer-Assisted Report-
ing; various U.S. government websites; various state and
local government websites; the websites of government
agencies worldwide.
23
Crowdlearning
Second year on the list
TREND 13
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
Searches for “what is the eu” and “what is brexit”
surged after the U.K. election.

Key Insight
In short, an adversarial piece of content—a photo, a video,
an audio file— is encoded with a tiny modification, usually
one that’s imperceptible to humans. It’s created in order to
help computer scientists adjust machine learning models.
Hackers use adversarial examples in a machine learning sys-
tem to attack it, causing the model to make a mistake.
Examples
In order for machine learning systems to learn, they must
recognize subtle differences. Researchers also use adver-
sarial information in order to train systems in how to recog-
nize misleading information in order to secure it. Adversarial
information is sort of like an optical illusion and it’s typically
imperceptible to the human eye or ear. It could be one pixel
out of a million that’s the wrong color or is misaligned—to
you, all those pixels together might still look like a photo of
a rainbow, but to a machine learning model, that one out-
of-place pixel could render the image gibberish. When that
happens, an adjustment is made to the system and it con-
tinues training.
What’s Next
Adversarial images can be used to knowingly and purpose-
ly trick a machine learning system. If an attacker trains a
model, using very slightly altered images, the adversarial
examples could then be deployed out into other models.
Adversarial examples can be embedded—intentionally, or
by accident—into photos, multimedia stories, virtual reality
content and the like. This is important to keep in mind, es-
pecially as fake news continues to proliferate in digital chan-
nels. It’s especially perplexing for search engines (Google,
Bing) and for any service that automatically tags our pho-
tos (law enforcement databases, Facebook).
Watchlist
Google’s Inception v3 algorithm and v4 algorithm; OpenAI;
EEECS at University of California-Berkeley; Stanford Univer-
sity; Kaggle competitions; Facebook; Microsoft; PRA Lab
at the University of Cagliari; University of Chicago; MIT’s
CSAIL; ImageNet database.
24
Adversarial Machine Learning
TREND 14
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
Extra information can be added to an image to fool
algorithms.

Key Insight
Computational photography is the convergence of comput-
er vision, computer graphics, the internet and photography.
Rather than relying on optical processes alone, it uses digi-
tal capturing and processing techniques to capture real life.
Examples
Everyone with a smartphone now has access to compu-
tational photography tools. In its iPhone 8 and iPhone X,
Apple uses computation photography to achieve a shallow
depth of field, while Facebook will soon automatically cor-
rect any 360-degree photos you upload.
What’s Next
New research from Nvidia and the University of Califor-
nia-Santa Barbara reveal a computational zoom technique,
which allows photographers to change the composition of
their photographs in real time. Photos are taken in a stack,
and then rendered with multiple views. This would allow
photographers to change perspective and the relative size
of objects within a photo after it has been taken. Other use
cases of computational photography include seamlessly
removing or adding objects to scenes, changing shadows
and reflections, and the like. Meanwhile, MIT’s CSAIL and
Google developed a technique that now automatically re-
touches and enhances the photos we take with our mobile
phones. Clearly there are ethical implications here for jour-
nalists—how much editing should be allowed and under
what circumstances? Likewise, journalists should develop
techniques to reveal how much editing has been done to
a photo—either intentionally or automatically—before using
them for reporting or in stories.
Watchlist
MIT’s CSAIL; MIT’s Media Lab; Nvidia; University of Cali-
fornia-Santa Barbara; Google; Apple; Samsung; Facebook;
Synopsys; Industrial Light and Magic; LG; Huawei; Morpho;
Qualcomm; Stanford University Computational Imaging
Lab; the Gcam team at Google Research.
25
Computational Photography
TREND 15
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
The Computational Zoom system makes it possible
to automatically combine wide-angle and telephoto
perspectives into a single multi-perspective image.
Image Credit: UCSB Mirage Lab.

Key Insight
You’ve no doubt heard of a bot: a software application that’s
been designed to automate certain tasks, such as scheduling or
managing basic customer service requests. In the past year, bots
have emerged from the fringe and have started to enter our main-
stream vocabulary. There are now more than 30,000 Facebook
Messenger. Beyond Facebook, Slack offers a number of produc-
tivity bots, while services such as Pandorabots allows developers
to deploy a chatbot across many platforms.
Examples
In the 1960s, Joseph Weizenbaum wrote a computer program
called Eliza that was capable of simulating a conversation be-
tween a psychiatrist and patient. It offered up plausible responses
to common questions. Today, newsrooms are now offering the
same basic system: offering plausible responses to questions
about the recent news events. Many newsrooms have experiment-
ed with chatbots, including BuzzFeed, TexasTribune, Quartz, Mic,
Los Angeles Times and elsewhere.
What’s Next
In March 2016, the world watched as @Tai.ai, a Microsoft exper-
imental Twitterbot, went on an anti-Semitic, homophobic, racist
rampage within 24 hours after its first tweet. Tai.ai was built on
the same platform as Microsoft’s experimental Mandarin-lan-
guage bot, Xiaoice. Both were capable of intimate conversations
with users, because the program is able to remember details
from previous conversations and because it mined the Internet
for human conversations in order to synthesize chat sessions. In
the summer of 2017, two of China-based Tencent’s bots—BabyQ,
co-developed with Turing Robot, and XiaoBing, co-developed
by Microsoft—went rogue the summer of 2017. During the recent
campaign cycle, we witnessed the rise of botnets—networks of
computers designed to send out spam. Fake social media ac-
counts, many of which originated in Russia, artfully tricked people
into having arguments about everything from Donald Trump to
immigration to taxes.
As we transition from text-based chatbots to voice interfaces,
newsrooms will need to determine how to interact with news con-
sumers. Some of the most interesting experimentation is coming
from China, which has hundreds of millions of users—their data is
helping to refine and recalibrate machine learning systems. Even if
a newsroom doesn’t deploy a chatbot, now is a good time to learn
and to develop strategies for audience engagement and revenue.
The groundwork for voice interfaces is being created from our
typed conversations today.
Watchlist
Chatfuel; Pandorabots; Twilio; Amazon; Facebook; iFlytek; Slack;
WeChat; Tencent; Baidu; Weibo; Alibaba; IBM; Alphabet; Micro-
soft; Snapchat; Coral Project.
26
Bots
TREND 16
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
The XiaoBing chatbot learned to hate the Communist
Party.

We are training bots in our own image. When developing your news bots in 2018, we recommend that you rate your work before
launch. Use this scale to rate your bot on its effectiveness—and to determine, in advance, whether or not you’ve accidentally encod-
ed bias into your system. Rate your bot on a scale of 0 - 10, with 10 being the highest (and preferred) score.
27
The Botness Scale
Does your bot reflect the values of your news-
room? How do you know for sure?
Is your bot’s purpose explicit? Will people inter-
acting with your bot clearly understand what its
purpose is after the first few interactions?
Does your bot perform its designated function
well?
Is your bot intuitive and easy to use, either on
a designated platform or across platforms?
Does your bot clearly explain where its answers
are coming from? Are you able to include any
evidence of your reporting, quotes and data?
Does your bot help people learn something new,
or does it effectively reinforce something that
people already know?
Does the corpus (the initial, base set of ques-
tions and answers) you’ve created reflect only
one gender, race or ethnicity? Or only one side
of a story? If so, was that intentional?
Did you assign your bot a traditional gender, eth-
nic or racial identity? If so, does it reference any
stereotypes?
Does your bot respond to gendered or sexist re-
marks? Does it respond to racial epithets or reli-
gious slurs? If it does respond, are the responses
appropriate to people of the group targeted?
Does your bot help people learn about their own
biases or broaden their worldviews?
01
02
03
04
05
06
07
08
09
10

Key Insight
We are entering an era of conversational interfaces. You
can be expected to talk to machines for the rest of your
life. These systems use semantic and natural language pro-
cessing, along with our data, in order to anticipate what we
want or need to do next.
Examples
If you’ve ever used Siri, Google Now, Amazon’s Alexa or
even the microphone button on your Comcast remote con-
trol, you’re familiar with voice interfaces. Soon, you will find
yourself talking to a host of connected devices, such as
your home thermostat, your car, your refrigerator, your ear-
buds, even your connected water bottle. By 2023, 50% of
the interactions North Americans have with machines will
be using their voices.
Conversational interfaces can simulate the conversations
that a reporter might have with her editor, as she talks
through the facts of a story. IBM Watson’s various APIs, in-
cluding Visual Recognition, AlchemyLanguage, Conversa-
tion and Tone Analyzer can all be used to assist reporters
with their work.
What’s Next
Amazon’s Alexa is quickly rising to become the default
platform for voice, with thousands of companies now in-
tegrating Alexa with their own products and services. That
includes an unusual collaboration with Microsoft, whose
Cortana now opens Alexa. Meantime, there is emerging re-
search into using voice interfaces to help professionals un-
derstand different sides of an argument. IBM recently built
a prototype that allows the user to ask a question—such as
“do violent video games contribute to violent acts in the
real world”—and receive a spoken analysis. A system like
this could one day be an invaluable newsroom tool, allow-
ing reporters to hash out their reporting and analysis with a
smart machine via a conversational interface.
Watchlist
Amazon; Alphabet; IBM Research; Cognitive Horizons Net-
work; Stanford University; MIT CSAIL; MIT Media Lab; Uni-
versity of Texas at Austin; Apple; Microsoft;
28
Voice Interfaces
Sixth year on the list
TREND 17
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
Amazon’s Echo is an voice interface found in many
American homes.

Key Insight
Our modern interfaces are becoming more and more like
ambient music—able to do more for us with fewer direct
actions, yet still be able to captivate our attention.
Examples
If you’ve interacted with Google Now or Amazon’s Alexa, or
if you own a Fitbit or Samsung Gear, of if you’ve gestured
to open your car’s trunk, you’ve used an ambient interface.
These are interfaces that automatically deliver information
or services, just as we need them. Devices with ambient in-
terfaces offer data, services and capabilities allow for com-
plex event processing, process management, automation of
information and tasks.
What’s Next
We are moving quickly in the direction of a post-screen fu-
ture. In our modern age of information, the average adult
now makes more than 20,000 decisions a day—and 226 of
them are about food alone2
. Emerging technology promises
to prioritize those decisions, delegate them on our behalf,
and even to autonomously answer for us, depending on the
circumstance. Much of this invisible decision-making will
happen without your direct supervision or input. Think of it
as a sort of autocomplete for intention. The power of am-
bient interfaces is explained by Metcalfe’s Law, which says
that the value of a network is the square of the total number
of people using it. As more people become part of ambient
networks of information, the more use cases we’ll see in the
future.
Watchlist
Alphabet; GE; Intel; Nvidia; Bosch Group; Samsung; Ama-
zon; Apple; Microsoft; Spotify; IBM; Tencent; Baidu; Alibaba;
Sony; NTT.
29
Ambient Interfaces
TREND 18
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
The Samsung Gear watch is an example of a wearable,
ambient interface.
We are moving quickly towards a post-
screen future.
2
According to researchers at Cornell University (Wansink and Sobal, 2007)

Key Insight
Deep linking has been around since the beginning of smart-
phones. Deep mobile links make it easier to find and share
data across all of the apps in your phone. They’re like hy-
perlinks, but rather than sending users to a web page, they
instead send users to specific screens in other mobile ap-
plications.
Examples
There are three kinds of deep links: traditional, deferred and
contextual. Traditional deep links reroute you from one app
or site (such as a link posted in Twitter) directly to the app,
as long as you have that app installed. Deferred deep links
either link straight to content if the app is installed, or to
an app store for you to download the app first. Contextual
deep links offer much more robust information—they take
you from site to app, app to site, or app to app, and they
can also offer personalized information. For example, when
you land at the airport, you might find that your airline app
sends you a link to Uber. (You’ll find similar offerings with-
in Google Maps.) Many of the new improvements to Ap-
ple’s iOS11 are built on deep linking: it allows users to easily
search through files and content, toggle between messag-
ing and apps, and get to content delivered by Siri.
What’s Next
This interoperability signals a new shift in thinking, as many
mobile app developers have been hesitant to use deep links.
With updates to Android and Apple, app-to-app experienc-
es should start to become more common. Deep linking is
vitally important for news organizations, as it is a way to
keep users within a news organization’s app.
Watchlist
Apple; Android; Facebook; Google; Bing; Appsfire; Branch;
Nielsen; Deeplink; MobileDeepLinking.org; Tencent; Alibaba;
AppsFlyer; Kochava; Tune; Adjust; Pinterest; Button; Yozio;
Baidu; AdRoll; Tapstream
30
Deep Linking
Fourth year on the list (non-consecutive)
TREND 19
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
Branch uses deep links to direct consumers from
social media feeds to products.

Key Insight
Productivity Bots will help journalists perform tasks more
efficiently.
Examples
Slack, with its over 5 million daily active users, and other en-
terprise messaging platforms like Yammer, HipChat, Ryver,
Zoom and Skype are becoming an increasingly important
component of the modern workplace as they begin to re-
place traditional productivity channels. Half of Slack users
reported a decline in email volume and a quarter reported
a similar decrease in in-person meetings. These platforms
include AI-powered bots that can help automate simple
tasks: scheduling editorial meetings, product workflows,
tracking and logging work.
What’s Next
As the developer ecosystems around these platforms con-
tinue to grow, new productivity bots will continue to drive
efficiency by helping these tools talk to each other. Bots are
getting smarter. With billions of messages sent daily across
a variety of chat applications, bots are being trained to lis-
ten to our conversations and pick up on when to jump in
and offer assistance. We expect this trend to continue as
new and better productivity bots will continue to emerge
and grow in their significance.
Watchlist
Slack; Alphabet; Yammer; HipChat; Ryver; Skype; Trello;
Dropbox; IFTTT; Heroku; Mailchimp; Zendesk; Microsoft.
31
Productivity Bots
TREND 20
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
The Ryver platform includes bot assistants for pro-
ductivity.

Key Insight
Artificial Intelligence is causing a disruption in education.
The “one size fits some” model will soon be replaced by
individualized adaptive learning software. This technology
can be used to help train newsrooms on the skills they’ll
need in the near-future.
Examples
Any good teacher is trained to pick up on signals from
students to drive their instruction. As software begins to
play an increasing role in the instructional delivery model,
these systems are being trained to do the same thing. Ma-
chine learning techniques powering the software requires
a large amount of data—which means many thousands of
students—to be effective. Online learning platforms such as
Khan Academy, EdX, Udemy and Coursera all use elements
of adaptive learning in their approach. Similarly, traditional
publishers like Pearson, McGraw-Hill, Cengage and others
are all developing online learning platforms that will incor-
porate varying adaptive elements. Education startups like
Acrobatiq, Cerego, and CogBooks all rely on adaptive sys-
tems to create individualized experiences for every student.
What’s Next
When newsrooms face revenue challenges, professional de-
velopment is often cut. Adaptive learning systems will be an
effective alternative to in-person newsroom training, lead-
ership development and workshops. As more and more in-
stitutions develop hybrid and online programs, and as more
students turn to alternative educational platforms, vast
amounts of data will be generated about their relative effi-
cacy. This will help determine exactly when adaptive learn-
ing is most effective, and when it is not, which will drive
innovation from startups and legacy publishers alike. How-
ever, proving efficacy in educational tools can often take
years, if it can be proven at all. The obvious benefits of on-
line, adaptive systems (easy to use, cost effective, individu-
alized) need to be weighed against the potential downsides
(reduced interactions with the instructor, focus on answers
instead of processes) before widespread adoption will take
root.
Watchlist
Acrobatiq; Cerego; CogBooks; Khan Academy; EdX; Ude-
my; Coursera; Pearson; McGraw-Hill; Cengage; Arizona
State University
32
Adaptive Learning
TREND 21
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
Adaptive learning software is being used to enhance
training and digital classroom instruction.

Key Insight
As the pace of technology adoption in the workplace contin-
ues to increase, the need for modern professionals to constant-
ly adapt to new platforms and learn new skills is becoming
paramount to their career development. Journalism is no ex-
ception. To help facilitate the goal of creating lifelong learners,
platforms like Udacity are developing nanodegree courses in
specific niche subjects to help individuals learn new skills and,
perhaps more importantly, to confer legitimacy in the eyes of
their prospective employers.
Examples
One theory emerging from Silicon Valley is that our traditional,
four-year post-secondary degree system alone cannot serve
our future workforce in the years to come. Human resources
directors and senior management are starting to see educa-
tion as a product, and they’re trying to maximize the ROE: Re-
turn-on-Education.
With the advent of automation and AI, journalists will need
highly-specialized skills, the sort that aren’t yet offered within
universities. Nanodegree provider Udacity has partnered with
universities such as San Jose State University and corpora-
tions like Alphabet, Facebook and ATT to create programs
for employees, to varying degrees of success. SJSU, for ex-
ample, suspended its partnership after more than half of the
students failed their final exams. In 2016, Udacity revealed a
new program called Nanodegree Plus, which guarantees stu-
dents a job within 6 months of graduation or it will refund tui-
tion. This is likely in response to several offline coding schools
like Flatiron School and Galvanize, which have offered similar
money-back guarantees to their graduates.
What’s Next
News organizations, journalism associations and professional
training groups should consider offering technical nanode-
grees as well as nanodegrees in newsroom leadership and var-
ious business skills. We expect to see continued consolidation
and scale in the maturing online and offline nanodegree mar-
ket, which should lead to some of the larger corporate and
university players coalescing around the winners. As some of
the players in the crowded coding bootcamp market have ei-
ther been consolidated or downsized, the money-back guar-
antee model has begun to come under fire. Is it sustainable to
guarantee employment to all your graduates within a certain
timeframe? For how long, and for which degrees?
Watchlist
Stanford University; MIT; Alphabet; Facebook; EdX; Coursera;
Udacity; Flatiron School; Galvanize.
33
Nanodegrees
TREND 22
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
Udacity offers nanodegree programs to help employ-
ees or job-seekers develop new skills to improve their
careers.

Key Insight
New technologies can be programmed to push or receive
information to/ from our mobile devices—and also our bod-
ies—tethering us to an always-on ubiquitous information
network.
Examples
You’ve no doubt heard about beacons, which are tiny devic-
es that can be programmed to push (or receive) information
to/from mobile phones using Bluetooth. We are located on
nearby networks, as sensors use our personal information
and collect data about our experiences. Beacons become
aware once you’re near them. They’re used frequently for
marketing, however they can also be used during planned
news/ culture/ arts/ sports events and throughout cities to
share news content with nearby people. Think of it as prox-
imity news.
What’s Next
Apple’s Fall Safari Technology Preview release (#38) ena-
bled the Beacon API by default and turned on beacon fea-
tures for iOS devices. Soon, we will be able to deliver prox-
imity-based news via WiFi, which can now identify you just
by bouncing signals around—your unique shape and pos-
ture are used to reveal who you are, even in a crowded room
of people. Emerging research has shown that WiFi can be
used to recognize what a person is saying or writing with
a pen—simply by analyzing the WiFi signals altered by our
bodies. In a confined space, like a conference center, sport-
ing arena or airport, this would allow a news organization
to recognize one of its news consumers and deliver stories
just for her.
Watchlist
Google’s Eddystone platform; Apple’s iBeacon platform; In-
doorAtlas; Unacast; Facebook; Blis; Snapchat; Polytechni-
cal University (China); MIT; University of New South Wales
(Australia); Oxford University; BLIP Systems; Bluedot; Gim-
bal; Qualcomm; Intel; Amazon.
34
Proximity News
Fifth year on the list
TREND 23
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
Proximity networks are being built for content distri-
bution.

Key Insight
Emerging predictive analytics tools wrangle your data, be-
havior and preferences in order to map your personality—
and predict how you’re likely to react in just about any situ-
ation. These tools can be used in journalism, to personalize
customer interactions and even to personalize the news
itself.
Examples
IBM Watson and Twitter offer a tool that mines Twitter
feeds and weather data to identify consumers who are like-
ly to fire off angry tweets if their cable service is disrupted.
Those complaints aren’t empty threats: IBM’s data shows a
correlation between disgruntled tweets and customer loss.
IBM’s technology can scan individuals’ social media data
and analyze their personalities to predict responses to an
email or an ad. Recruiting startups, dating sites and school
application platforms are all starting to experiment with
personality recognition software. Nashville-based startup
Crystal culls thousands of public data sources to help you
learn about someone’s personality before calling or email-
ing them. It even offers a kind of spell check for sentiment,
autocorrecting phrases and making recommendations
(“keep the message under 200 words, otherwise this recip-
ient might ignore it”) so that the message resonates better
with your intended recipient.
What’s Next
These tools can be used to provide better customer inter-
actions for news consumers: content could be personalized
and targeted to specific individuals. Personality recognition
can also be used, along with natural language generation
algorithms, to personalize parts of stories to make them
more relatable to individual readers.
Also on the horizon is facial and tonal recognition. Facial
and voice recognition analytics will help machine learning
systems to detect consumers’ emotional state in real-time.
Mattersight Corporation is using personality and behavior
to route calls through call centers, and its latest “Predictive
Video” system promises to analyze your speech and facial
expressions from any video where you’ve appeared.
Watchlist
Mattersight Corporation; MIT; IBM; Twitter; Crystal; Stanford
University; Salesforce; Autodesk; Symantec; Mobileye; Intu-
it; Adobe.
35
Personality Recognition
and Analytics
TREND 24
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
Personality recognition can also be used, along with
natural language generation algorithms, to person-
alize parts of stories to make them more relatable to
individual readers.

Key Insight
In this modern digital age, attention is currency. As tech-
nology has evolved, news organizations have adapted their
existing content for the screens of our ever-changing de-
vices. However multiple studies show that our attention is
continually split between what we’re doing in the real world
and what we’d like to be doing online. As the two become
more intertwined, capturing our attention is becoming
more difficult than ever.
Examples
While the 2016 election season helped grow the audiences
of news organizations, it also brought alternate sources of
information, splintering the attention of consumers across
quality and questionable news. Making sure that content fits
correctly on a screen is only solving part of the challenge—
what about content fitting our needs and behaviors as both
change throughout the day? In order to capture someone’s
attention, you must consider a number of variables: where
is she right now? What’s she likely to be doing in the next
60 seconds? What’s relevant to her in the next few min-
utes? What need can you fulfill for her at this moment?
Attention is an increasingly important metric for advertis-
ers, media buyers and ad exchanges, so there is a finan-
cial incentive for news organizations to shift their strategic
thinking. There has been tremendous consolidation in the
measurement and online advertising space as well, espe-
cially by IBM, Google, Facebook, Quantcast and Adobe.
What’s Next
Going forward, every news organization must focus more
of its attention on the consumer herself and what she is
doing. Soon, journalists will work alongside algorithms to
syndicate different versions to different devices depend-
ing on a user’s individual needs, given that those needs will
change throughout her day.
Watchlist
Omniture; Nielsen; comScore; Facebook; Chartbeat; Simp-
li.fi; Adobe; Quantcast; The Media Trust; Visible Measures;
IBM; Facebook; Chartbeat; Google.
36
Attention
Fourth year on the list
TREND 25
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
The Future Today Institute’s Attention Matrix is a tool
to help measure whether your strategy will command
the attention of your desired audience.

Key Insight
In the past three years, we’ve seen the first widespread
cases of important journalism being erased from the web
because of media consolidation or because sites were no
longer being maintained. Digital Frailty is the phenomenon
in which those digital assets published to a news organiza-
tion’s website are impermanent or easily broken.
Examples
Perhaps not every Facebook post should be saved in per-
petuity, but might we need to look back on this moment in
time and reflect on how our language—how the very way
we communicate—was shaped by our Instas, our Snaps,
and our tweets? Will our future historians look back, marve-
ling at the amount of anthropological data we were simul-
taneously creating—and destroying? If this past election
season taught us anything, it’s that Twitter helped to shape
public opinion and the outcome of the election, even as
many controversial tweets posted by candidates running
for office, were deleted by their campaigns.
A Pulitzer Prize-winning investigative series about a col-
lision that killed 20 children and devastated a Colorado
community went offline when the Rocky Mountain News
went out of business. The Tampa Tribune, whose motto was
“Life. Printed Daily,” kept its rival, the Tribune, hunting for
important stories in the public interest, covering investiga-
tions into Tampa’s judges, legislators and law enforcement.
Humanity operates on a continuum. After devastating
Texas, Hurricane Harvey made landfall near New Orleans
on the 12th anniversary of Katrina. Rising From Ruin, an
award-winning project by MSNBC, told the Katrina’s after-
math through the lenses of two small communities in Mis-
sissippi that weren’t covered by any other media outlet. It
included a series of videos, maps, interactive elements, a
forum for residents—and since it only existed as a website,
there was no other way to see the stories. When Microsoft
pulled out of its joint venture with NBC, the project went
offline.
Digital Frailty in Government and Public Information
American journalists watched as U.S. government agencies
removed studies, data and reports throughout 2016 and
2017. Most notably, the Environmental Protection Agency
scrubbed its website of climate change information. This
was an effort to support the Trump Administration’s ideas
and policies. A government website built to educate chil-
dren, called “Energy Kids,” also scrubbed mentions of cli-
37
Digital Frailty
TREND 26
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
A screenshot of the U.S. Environmental Protection
Agency’s website taken on September 6, 2017.

mate change. The Trump Administration also removed LG-
BTQ content from federal websites, scrubbed a lot of civil
rights information off of WhiteHouse.gov and scrubbed
the HHS.gov website of healthcare data.
What’s Next
This is a phenomenon affecting journalists everywhere.
Digital frailty isn’t just about falling revenue—sometimes,
new technology obviates the old, before anyone’s had a
chance to convert files or develop archives. News execu-
tive Mario Tedeschini-Lalli explains how Italy’s largest news
website, Repubblica.it, didn’t originally use a content man-
agement system. When the site installed a CMS for the first
time, everything published before it was lost forever. While
some content can be retrieved via the Internet Archive, it is
only taking snapshots of content at a time. Libraries archive
printed material, but there is no central repository for all
of the digital content we are now producing. Perhaps we
don’t need to save every listicle and quiz. What will a future
society look like if our current media landscape goes dark?
Do we have an obligation to preserve the digital conver-
sations shaping society? Should we be working harder to
ensure that digital archives aren’t lost?
Watchlist
Axel Springer; Yahoo; Tumblr; Hearst Corporation; Time Inc;
Yomiuri Shimbun Holdings; Tronc; Gannett; Viacom; Hubert
Burda Media; Comcast; Alphabet; Asahi Shimbun Compa-
ny; Microsoft; Grupo Globo; Advance Publications; News
Corp; Univision; Baidu; Bertelsmann; Twitter; Snap; Insta-
gram; General Electric; Bloomberg; Disney; Amazon; ATT;
Verizon; ESPN; Netflix; Hulu; The Onion; PRX; PRI; Internet
Archive; news organizations everywhere.
38
Digital Frailty cont.
TREND 26
If a Pulitzer-finalist 34-part series of
investigative journalism can vanish from
the web, anything can.
- Adrienne Lafrance

Key Insight
In the past year, credible news organizations have faced
a crisis of confidence caused by Twitter bots, political ex-
tremists, and elected officials. Radical transparency offers
the public a full view of how the story was reported and
produced.
Examples
There are too many instances of “fake news” accusations
to list. In order for journalists to combat a growing, but un-
founded, public distrust, they should offer radically trans-
parent reporting. PolitiFact, the Pulitzer Prize-winning
fact-checking website, lists all of the sources used for a
story. ProPublica’s /nerds blog explains some of the work
behind data journalists, developers and reporters’ stories.
What’s Next
Professor Ahmed Elgammal at Rutgers University devel-
oped an algorithm that looks for novelty in paintings and
analyzes which artists influenced that work. His research
has inspired others to use similar network analysis, histori-
cal data and machine learning to look for similarities in lit-
erature, writing and news. A system like this could be de-
ployed to look for explicit and hidden influencers on news
stories. Now that news organizations are relying on data,
algorithms, and machine learning for various aspects of
news gathering and publishing, they should commit to rad-
ical transparency. There are too many instances of bias in
algorithms to list. Just as consumers expect to see a byline
on stories, because it creates a chain of accountability, they
will soon expect to know how stories were built. Report-
ers aided and augmented by smart systems should explain
what data sets and tools they used. Meanwhile, stories that
were written in part or entirely by computers should reflect
that an algorithm was responsible for the piece of content
being read/ watched.
Watchlist
News organizations everywhere.
39
Radical Transparency
TREND 27
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
In this age of technology, we need a nutritional label
for news.

Key Insight
Some organizations have begun to experiment with tem-
porary products: limited-run newsletters, podcasts that
only last a set number of episodes, live SMS offerings that
happen only during events.
Examples
News organizations creating limited-edition news prod-
ucts, do not necessarily need to create many labor-inten-
sive, one-off templates and workflows. Producers can de-
velop templates that can be iterated on and redeployed
again. BuzzFeed stood up a temporary chatbot during the
political conventions in 2016, while the New York Times
launched a short-term chat service for the Olympics.
Whether it’s a planned news event (such as local elections,
festivals or races), an annual conference (ONA, SXSW,
PopTech), a season (skiing, football, baseball), or a big sto-
ry that has a defined beginning middle and end (such as a
weather event), limited-edition news products are started
to be used by news organizations.
What’s Next
We anticipate seeing more temporary podcasts, newslet-
ters and chatbots that are deployed specifically for just one
event. Limited-edition news products are revenue and au-
dience engagement opportunities, as they are vehicles for
data collection and targeted advertising.
Watchlist
News organizations everywhere.
40
Limited-Edition News Products
TREND 28
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
BuzzFeed’s BuzzBot was active during the 2016 Re-
publican National Convention.

Key Insight
Newsletters, podcasts and niche networks that captivate
smaller audiences made a huge comeback between 2015-
17. What’s next is an expansion to capture even more niche
audiences.
Examples
Suddenly, it seems like everyone—from world leaders, to
your next-door neighbor—has a podcast, newsletter, a chat-
bot or all three. This is due in part to services like Mailchimp,
TinyLetter (owned by Mailchimp), Skype, Google Hang-
outs, Garage Band, SoundCloud, Libsyn, Stitcher, Auphon-
ic, SpeakPipe and a host of affordable smartphone micro-
phone attachments. In 2017, new niche media empires took
root: Jessica Lessin’s The Information publishes in-depth
stories on tech and business. Former MTV chief digital of-
ficer Jason Hirschhorn expanded his REDEF newsletter
empire.
What’s Next
Our research indicates that more niche networks will con-
tinue to launch with content distributed in myriad formats.
We also expect to see more niche-focused digital-only
content products—private content networks, short-form
podcasts, and augmented reality integrations—in 2018 and
2019. Smaller sites like, Nautil.us, Pacific Standard, Bitter
Southerner, New Inquiry and Aeon produce exceptional
content and command very attentive audiences. Our re-
search shows that there is profit to be made, even though
audiences may be smaller in size. As many of the one-to-
few startups have proven in the past 24 months, an influen-
tial network with sticky engagement shows why dedicated
attention matters more than a bunch of clicks, and that’s
the metric that will matter most in the near future. Adver-
tisers are taking notice.
Watchlist
REDEF Group; The Information; PRX; TinyLetter; Mailchimp;
Nautilus; Pacific Standard; Bitter Southerner; New Inquiry;
Aeon; Backchannel; Skype; Garage Band; SoundCloud; Lib-
syn; Stitcher; Auphonic; SpeakPipe; Twilio; PRI.
41
One-To-Few Publishing
TREND 29
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
Nautil.us is a new breed of website with a highly en-
gaged niche audience that pays for content.

Key Insight
Notifications show bits of information, including updates,
reminders and messages from friends. They appear on the
lock screens of mobile phones, wearables and connected
devices.
Examples
Notifications are particularly attractive to news organiza-
tions because they capture attention when our attention
is most vulnerable. Leveraging our FOMO (fear of missing
out), notifications tempt us to look at our screens and to
click through. Users who opt-in to receive push notifica-
tions increase app retention rates by 2x or more. Opt-in
users are twice as likely to engage with the content teased.
Most major news organizations, as well as content-creators
from other sectors, are now engaging notifications to pull
users into content.
What’s Next
The problem is that notifications now come from every-
where—from the OS, government emergency services,
weather apps, games, social networks, podcasts, and
more. Notifications with photos and emoji perform better,
which is a show of how cluttered the space has become.
News organizations will need to develop new tactics and
strategies to ensure that their notifications don’t add to
the existing notification layer clutter—and so they do not
alienate readers.
Watchlist
News organizations everywhere; Android; Apple; Amazon;
Microsoft.
42
Notification Layer
TREND 30
Notification screens are prized real estate.
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch

Key Insight
On the fringes, news organizations are beginning to pro-
vide journalism as a service, rather than solely distributing
traditional news products.
Examples
“Software as a Service” is a licensing and delivery model,
where users pay for on-demand access. It’s a model that in
the near-future might be an inevitability. The central chal-
lenge within news organizations is that there are immedi-
ate, acute problems—but reasonable solutions will require
long-term investment in energy and capital. The tension
between the two always results in short-term fixes, like
swapping out micro-paywalls for site-wide paywalls. In a
sense, this is analogous to making interest-only payments
on a loan, without paying down the principal. Failing to
pay down the principal means that debt—that problem—
sticks around longer. It doesn’t ever go away. Transitioning
to “Journalism as a Service” enables news organizations to
fully realize their value to everyone working in the knowl-
edge economy—universities, legal startups, data science
companies, businesses, hospitals, and even big tech giants.
News organizations that archive their content are sitting on
an enormous corpus—data that can be structured, cleaned
and used by numerous other groups.
What’s Next
News deployed as a service includes different kinds of par-
cels: news stories; APIs; databases that can be used by both
the newsroom and paying third parties; calendar plug-ins
for upcoming news events; systems that can automatical-
ly generate reports using the news org’s archives and da-
tabases and the like. Services work outside of the social
media landscape, relieving news organizations of revenue
sharing and allowing them to fully monetize their services.
Watchlist
PRX; Twilio; REDEF Group; The Information; The Coral Pro-
ject; MIT Media Lab; ProPublica.
43
Journalism as a Service (JaaS)
TREND 31
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
News organizations will find new ways to generate
revenue through Journalism as a Service.

Key Insight
Social networks are under pressure to offer more transpar-
ency in the numbers they report back to news organiza-
tions While most companies that publish content on the
web are obsessed with metrics, historically they’ve kept au-
dience data hidden from staff.
Examples
Metrics are neither easy to find nor easy to understand for
many working inside of content organizations. Facebook
has apologized for misreporting its metrics, which includ-
ed displaying incorrect numbers of video plays to adver-
tisers and publishers. The company said that it had been
showing incorrect metrics for two years as it attempted to
challenge YouTube. Earlier in the year, current and former
Facebook staff alleged they were instructed to suppress
conservative news from the site’s “Trending Topics” area.
During the summer of 2017, Facebook offered new landing
page views and page interaction metrics, which the com-
pany said would offer better insights for advertisers.
It goes without saying that metrics can influence editorial
and business decisions, not to mention how the public in-
terprets the popularity of a story. Most large news organ-
izations have hired audience engagement and analytics
managers as go-betweens.
What’s Next
Publishers and advertisers will question the validity of met-
rics that they, themselves, cannot verify. Anyone creating
content needs to understand the ebb and flow of traffic
and how one piece of content fits into the broader scope
of the organization. We also expect to see news and other
content organizations develop new models to bring trans-
parency in metrics to staff—without jeopardizing editorial
integrity.
Watchlist
Nielsen; Chartbeat; YouTube; Google; Instagram; Snap;
Facebook; Twitter; news organizations everywhere.
44
Transparency in Metrics
TREND 32
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
Historically, news organizations have kept audience
data hidden from staff, while third-party services
haven’t always been transparent about what numbers
they’re counting.
Photo Credit: http://www.adoraattack.com/
fuzzy-numbers/

Key Insight
Buoyed by charges of “fake news,” real-time fact-checking
will be a priority for journalists in 2018.
Examples
Digital tools have made it easy to report on a live event and
publish in real time, but adding context—such as whether
or not a source’s statement is factually accurate—usually
happens after. In 2016, the presidential debates were fact
checked by a number of groups, including National Public
Radio (NPR), the Washington Post, and even Hillary Clin-
ton’s own staff. The efforts were people-powered. In Feb-
ruary 2017, Washington Post reporters fact checked Presi-
dent Trump’s address to Congress with very little lag.
What’s Next
Late in 2016, Google introduced a fact-check tag to its
Google News service—readers can see fact checks next to
trending stories. As we now see on a near-daily basis, in-
accuracies and falsehoods quickly spread on social media
masquerading as the truth. At least when it comes to cit-
ing numbers and data, artificial intelligence will soon allow
news organizations to automate the fact checking process.
In a few years, AI systems will enable more sophisticated
fact checking: explaining whether information was taken
out of context, or exaggerated, or downplayed.
Our analysis indicates that news organizations will soon
have a tremendous opportunity to use AI along with social
media data and their own article databases, to build tools
for real-time fact checking, adding a critical editorial layer
that’s both good for the public interest and good for build-
ing brand reputation.
Watchlist
IBM Watson; Tencent; Baidu; Google; Amazon; Facebook;
Twitter; news organizations everywhere.
45
Real-Time Fact Checking
TREND 33
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
The Washington Post has been experimenting with
faster fact-checking.

Key Insight
As consumers shift to their mobile devices, developers are
making sure their apps work offline.
Examples
In the U.S., consumers now spend an average of five hours
a day on their mobile devices. As consumers move about
our days—commuting, walking around the office, or sitting
through a Little League game—they still find themselves
offline. A number of news aggregators—including Google,
Smartnews and Apple—want to capitalize on the time con-
sumers devote to their screens, even when the WiFi signal
is weak. The Washington Post’s progressive web app cuts
mobile page load times from 4 seconds to 80 milliseconds
and allows consumers to read news stories without a data
or WiFi connection.
What’s Next
Until news consumers have ubiquitous access to cheap, fast
data, offline reading will be a necessity. News organizations
that include seamless, offline experiences will find stickier
audiences.
Watchlist
Tencent; Baidu; Google Play; Pocket; Amazon; news organ-
izations everywhere.
46
Offline Is The New Online
TREND 34
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
New techniques allow consumers to access news con-
tent, even when they’re not on a strong network.

Key Insight
As news organizations venture into podcasts, new search
tools allow the newsroom—as well as news consumers—to
find exactly what information they’re looking for within au-
dio-only content.
Examples
While developers have learned how to quickly index and
display web content, digital audio has always remained an
unsolved challenge. Now, rather than searching for a top-
ic and getting a bunch of hyperlinks to click through and
listen to, consumers will instead receive a series of buttons
that play the exact snippet of audio that’s related to their
search. Better than buttons, consumers can also speak their
searches to a voice assistant and immediately get to the
podcast they were trying to remember, to replay a news
report they’d heard in the car, or to get a series of clips re-
lated to a subject they’re interested in.
Startup Audioburst uses artificial intelligence to index au-
dio broadcasts and make them easier for consumers to
find. Rather than searching for keywords, Audioburst uses
natural language processing to automatically discover the
meaning conveyed and to surface the right content. For ex-
ample, if a consumer wants an update on how close the U.S.
is to a conflict with North Korea, she can ask a voice-acti-
vated app (Amazon’s Alexa, Google Home), which will sift
through audio information and deliver a set of clips.
What’s Next
With so much funding and development into voice inter-
faces, audio search will quickly become one of the most
important tech trends in the years to come.
Watchlist
Audioburst; Amazon; Google; Apple; Advanced Media; Vi-
acom.
47
Audio Search Engines
First year on the list
TREND 35
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
Audioburst uses artificial intelligence to index audio
broadcasts and make them easier for consumers to
find.

Key Insight
Entrepreneurs are building and preparing to launch thou-
sands of low-cost, high-value satellites in the next three
years. These satellites are small, capable of communicating
with each other, and will photograph every inch of Earth’s
surface every day of the year.
Examples
Miniature satellites, otherwise known as CubeSats, aren’t
new technology. They’ve actually been in use by space
agencies for years. What’s changing is the launch technol-
ogy that lifts CubeSats into orbit. Heavy investment into
propulsion systems—not to mention significant advance-
ments in technology and cheaper components—are mak-
ing it easier to mass-produce tiny satellites in a factory and
launch them for a variety of purposes. Fleets of CubeSats
now take photos of farmland and beam them back down
to earth to help farmers assess their crops. Image analysis
software can tell big box retailers, such as Best Buy, how
many cars are parked in their lots and look for trends over
time. They can then do the same with a competitor’s park-
ing lots to gather strategic intelligence. Mining companies
can survey a swath of land to see who’s started drilling and
whether they’ve struck oil. Satellites monitor traffic, polar
ice caps, and even us. Unlike a traditional, large satellite,
when one CubeSats goes offline or gets damaged, the rest
of the fleet still works.
Near-real time images, coupled with machine learning and
analysis tools, is big business. Governments, big agricultur-
al corporations, intelligence agencies, shipping companies
and logistics firms all want access, so they’re willing to pay
tens of millions of dollars a year for access. The combined
valuation of companies such as Planet, Airbus DS, MDA
and DigitalGlobe is well into the tens of billions.
What’s Next
The Federal Aviation Administration is projecting “an un-
precedented number” of satellite launches between 2018-
2020. News organizations could gain access to the images
and tools for data-driven reporting projects and to under-
stand the world as it’s happening, in real time. CubeSats
and image analysis will help reporters take the pulse of their
cities, gain a deeper view into weather events and dive into
criminal activity.
48
CubeSats
TREND 36
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
CubeSats can be used alone or stacked to suit the
needs of a specific mission.
Credit: Canadian Space Agency

Watchlist
Space Systems Loral; MDA; Planet; Planetary Resources;
Airbus DS; DigitalGlobe; National Geospatial Intelligence
Agency; 3 Gimbals; Space Exploration Technologies Corp;
Orbital Insight; Google; SpaceKnow; Capella Space Inc;
OneWeb; SpacePharma; Santa Clara University; Technis-
che Universitat Berlin; Tokyo Institute of Technology; Uni-
versity of Tokyo; California Polytechnic University; Cornell
University; Boeing; Delft University of Technology; NASA
Ames Research Center; Transcelestial; NSLComm; Earth-
cube; Aerial Maritime; Fleet Space; Astrocast; Kepler
Communications; GeoOptics; Hera Systems; Sky and Space
Global; Astro Digital; Kanagawa University; The Aerospace
Corporation; Los Alamos National Lab; NRL Naval Center
for Space; Space and Missile Defense Command; Satellog-
ic; Spire; US Air Force; Lawrence Livermore National Labra-
tory; MIT; Shenzhen Aerospace Donganghong; National
University of Defense Technology (China); Shanghai Engi-
neering Center for Microsatellites (China); SRI International;
Naval Postgraduate School.
49
CubeSats cont.
TREND 36

Key Insight
U.S. adults now spend close to an hour a day watching on-
line video, and increasingly we’re using our mobile phones
to access that content. But not all adults prefer video. A
Pew Research Center survey3
found that more Americans
prefer to watch their news (46%) than to read it (35%) or
listen to it (17%). But the demographics might surprise you:
Americans age 50 or older prefer video, while the majority
of 18 to 29-year-olds (42%) prefer reading the news. Still,
advertising and marketing budgets are flowing freely to the
agencies creating video—and to the platforms distributing
it. Mobile video ad spending will reach $18 billion in 2018.
37 Connected TVs
TVs that connect to the internet certainly aren’t new. What’s
changed is penetration in average households and the availability
of streaming apps that bypass the standard list of cable and pub-
lic broadcasting channels, such as Amazon Prime Video, Roku,
Hulu, YouTube, Showtime Anytime, iPlayer (UK-only), All 4 (UK
only), Playstation Now, HBO Now, Direct Now, Plex, iTunes, and
of course, Netflix.
Impact on news organizations
Streaming services will erode local broadcast news mar-
kets. These services will also disrupt longer-form television
news broadcasts.
50
Video
Seventh year on the list
TRENDS 37 - 39
Informs
Strategy
Revisit
Later
Act
Now
Keep
Vigilant
Watch
The digital video ecosystem will continue to grow in
2018.
3
http://www.pewresearch.org/fact-tank/2016/10/06/younger-adults-more-likely-than-their-elders-to-prefer-reading-news/

Emerging digital trends for journalism

Emerging digital trends for journalism

Empfohlen

Empfohlen

Weitere ähnliche Inhalte

Was ist angesagt?

Was ist angesagt? (20)

Ähnlich wie Emerging digital trends for journalism

Ähnlich wie Emerging digital trends for journalism (20)

Mehr von Filipp Paster

Mehr von Filipp Paster (20)

Kürzlich hochgeladen

Kürzlich hochgeladen (20)

Emerging digital trends for journalism